On Mon, Feb 25, 2019 at 12:17:36PM +0100, Johannes Thumshirn wrote: > On 22/02/2019 15:55, David Sterba wrote: > >> > >> - return csum_tree_block(fs_info, eb, 0); > >> + if (WARN_ON(csum_tree_block(eb, result))) > > > > I think the warn should go to csum_tree_block when the mapping function > > returns 1, there's even a comment explaining why it can't normally > > happen. The reason for the warning in csum_dirty_buffer is not very > > clear from the context. > > > > Otherwise ok, the verify or write semantics of a function that (by name) > > only calculates checksum is confusing. Thanks. > > Sure. Would it make sense to split this into two patches? Yeah, makes sense. One to move the verify part and another to add the WARN_ON. Thanks.
